✅ Theoretical sampling, open sampling

❎ Convenience sampling, purposive sampling, selective sampling, snowball sampling

What is theoretical sampling?

Theoretical sampling ‘is controlled by the emerging theory’ and is ‘the process of data collection for generating theory’ (Glaser, 1978 p. 36). This simply means that as we analyse data, questions arise, gaps in the theory are noticed and we theoretically sample for particular data to address those questions and gaps. Note that the emphasis is on making theoretical decisions rather than logical decisions.

In a new study one has to start somewhere and any data in any form that is from the substantive area of interest is potentially useful. Glaser suggests that one might ‘begin by talking to the most knowledgeable people to get a line on relevancies and leads to track down more data and where and how to locate oneself for a rich supply of data.’ (Glaser, 1978 p. 45). Note the aim of getting a line on relevancies rather than finding data about ‘x‘.

Glaser and Strauss emphasise ‘Our criteria [for theoretical sampling] are those of theoretical purpose and relevance—not of structural circumstance. (1965, p. 48).

A gap in GT terminology?

GT texts advise us that theoretical sampling is conducted from day one of a grounded theory study. In practice, however, Glaser found that novice grounded theorists followed theoretical sampling advice too literally, especially the advice to use what they learned in their first research conversation to inform the questions asked in the second and so on. As a result novices started delimiting their study – narrowing down their focus – too quickly. At best this delays theory development and at worst compromises the quality of the theory developed.

Instead, in seminars, Glaser encouraged us – during the first, more inductive, ‘open coding’ stage – to remain open to the data in its collection and analysis until we had generated the problem and a potential core category. Then to theoretically sample for and ‘selectively code’ particular data in the second, more deductive, stage of analysis. 

For novice grounded theorists, I notice that there appears to be a sampling naming gap in the open coding stage into which convenience sampling, selective sampling, purposive sampling  and snowball sampling all rush in. 

Convenience sampling: ‘involves choosing the nearest and most convenient persons to act as respondents’ (Robson 2024, p.363).

Purposive sampling: ‘also known as judgmental, selective, or subjective sampling, is a non-probability sampling technique where researchers intentionally select participants based on specific characteristics relevant to their study’  (Bullard, 2024).

‘The logic on which you base your strategy for selecting cases for a purposive sample should be dependent on your research questions and objectives’ (Saunders et al, 2003 p. 175).

Snowball sampling: ‘Here the researcher identifies one or more individuals from the population of interest. After they have been interviewed, they are used as informants to identify [name] other members of the population, who are themselves used as informants, and so on’ (Robson 2024, p.363).

These sampling approaches tend to focus on how to identify participants logically as opposed to how to collect data of many different types theoretically.

Glaser resisted ‘fracturing’ the concept ‘theoretical sampling’ (Glaser, 1992 p. 102), however to address the emergent issue of a naming gap, I propose that we begin theoretical sampling with ‘open sampling’:

Open sampling is based on the twin tenets of ‘all is data’ and ’emergence’ (Glaser, 1992). The tenet of emergence insists that the boundaries of the study, the concepts and how they relate and integrate to form a theory all emerge during analysis. Together, these two principles mean that open sampling is conducted in as unstructured way as is feasible and that many forms of data can be collected as long as they are of the study’s substantive area of interest. (Scott, 2025, pp. 40 – 41. Emphasis added. )

What is the problem with logical sampling?

Logical sampling undermines the grounded theory research process. Grounded theory is an exploratory research method where we privilege the concerns of the research population over our own, so that we have the best chance of finding the strongest patterns: the theory we develop should fit the data we collect. 

In a verification/confirmation study, the researcher privileges their own perspective and anticipates the theory: they seek the data to fit the theory they have in mind.  When the researcher knows what they are seeking they lean towards logical sampling techniques. 

In a verification study one might be pursuing a specific hypothesis or idea – perhaps assuming stress as a problem for senior managers: ‘For example a group of principals and senior managers of secondary schools is chosen as the research is studying the incidence of stress amongst senior managers’ (Cohen et al, 2003, p. 103). This is in contrast to having an open, exploratory mindset when collecting data: for instance, my experience – my theoretical sensitivity – tells me that some senior managers relish the challenge and excitement of their work rather than experiencing it as stress, and so I would not assume stress. A mindset of verification/confirmation can lead to logical sampling and risks focusing on weaker patterns of less relevance to participants.

Instead of logical sampling to seek to verify or confirm an idea, the grounded theorist samples theoretically in order to generate a grounded theory. 

‘Theoretical sampling is not what Leonard Schatzman has aptly termed ‘selective sampling’, which is a frequently used sampling method in qualitative analysis. Selective sampling refers to the calculated decision to sample a specific locale according to a preconceived but ‘reasonable’ initial set of dimension, (such as time, space, identity or power) which are worked out in advance for a study. The analyst who uses theoretical sampling cannot know in advance precisely what to sample for and where it will lead him.’ (Glaser, 1978, p. 37)

On this page I explain why logical sampling is consistent with a verification/confirmation study and is therefore inappropriate for an exploratory grounded theory study. Instead I encourage you to begin theoretical sampling with open sampling. These issues are explored further in Using Grounded Theory: How to Develop Theory for Managed Change (Scott, 2025) especially Chapters 1 and 3.

Helen Scott PhD
Grounded Theory Online
September, 2025